| Metric | naive | lstm | xgboost | cnn |
|---|---|---|---|---|
| diff_of_means | -1.098 | -5.145 | 7.534 | -8.937 |
| ratio_of_sd | 0.830 | 0.933 | 0.857 | 0.877 |
| amplitude_ratio_of_means | 0.539 | 0.645 | 0.529 | 0.685 |
| maximum_error | 0.302 | 0.308 | 0.277 | 0.319 |
| ks_mean_on_coarse_res_with_extremes | 0.587 | 0.316 | 0.436 | 0.263 |
| rainy_hours_ratio_of_means | 0.815 | 0.861 | 0.951 | 0.872 |
| qqplot_mae | 0.029 | 0.018 | 0.018 | 0.032 |
| acf_mae | 0.152 | 0.096 | 0.110 | 0.085 |
| extremogram_mae | 0.109 | 0.056 | 0.083 | 0.060 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | lstm | xgboost | naive | cnn |
|---|---|---|---|---|
| diff_of_means | -3.851 | 4.573 | -4.781 | -6.865 |
| ratio_of_sd | 0.901 | 0.878 | 0.894 | 0.848 |
| amplitude_ratio_of_means | 0.609 | 0.526 | 0.552 | 0.653 |
| maximum_error | 0.314 | 0.270 | 0.316 | 0.319 |
| ks_mean_on_coarse_res_with_extremes | 0.240 | 0.387 | 0.378 | 0.186 |
| rainy_hours_ratio_of_means | 0.819 | 0.901 | 0.792 | 0.852 |
| qqplot_mae | 0.027 | 0.019 | 0.028 | 0.040 |
| acf_mae | 0.102 | 0.121 | 0.147 | 0.086 |
| extremogram_mae | 0.070 | 0.090 | 0.117 | 0.063 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | lstm | naive | xgboost | cnn |
|---|---|---|---|---|
| diff_of_means | -3.443 | -4.318 | 5.570 | -5.838 |
| ratio_of_sd | 0.920 | 0.901 | 0.890 | 0.866 |
| amplitude_ratio_of_means | 0.609 | 0.551 | 0.533 | 0.657 |
| maximum_error | 0.308 | 0.315 | 0.267 | 0.322 |
| ks_mean_on_coarse_res_with_extremes | 0.302 | 0.474 | 0.407 | 0.196 |
| rainy_hours_ratio_of_means | 0.849 | 0.808 | 0.937 | 0.875 |
| qqplot_mae | 0.025 | 0.026 | 0.017 | 0.037 |
| acf_mae | 0.101 | 0.154 | 0.117 | 0.082 |
| extremogram_mae | 0.074 | 0.124 | 0.094 | 0.063 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | lstm | cnn | xgboost | naive |
|---|---|---|---|---|
| diff_of_means | -10.220 | -10.327 | -18.592 | -24.890 |
| ratio_of_sd | 0.943 | 0.875 | 0.933 | 0.884 |
| amplitude_ratio_of_means | 0.685 | 0.717 | 0.618 | 0.599 |
| maximum_error | 0.335 | 0.315 | 0.285 | 0.295 |
| ks_mean_on_coarse_res_with_extremes | 0.244 | 0.176 | 0.340 | 0.442 |
| rainy_hours_ratio_of_means | 0.809 | 0.821 | 0.740 | 0.655 |
| qqplot_mae | 0.023 | 0.034 | 0.037 | 0.053 |
| acf_mae | 0.114 | 0.099 | 0.145 | 0.182 |
| extremogram_mae | 0.048 | 0.045 | 0.082 | 0.103 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | lstm | cnn | naive | xgboost |
|---|---|---|---|---|
| diff_of_means | -3.176 | -4.993 | 19.995 | 30.082 |
| ratio_of_sd | 0.899 | 0.851 | 0.709 | 0.707 |
| amplitude_ratio_of_means | 0.600 | 0.636 | 0.451 | 0.416 |
| maximum_error | 0.286 | 0.336 | 0.281 | 0.248 |
| ks_mean_on_coarse_res_with_extremes | 0.333 | 0.294 | 0.569 | 0.461 |
| rainy_hours_ratio_of_means | 0.926 | 0.953 | 0.995 | 1.204 |
| qqplot_mae | 0.025 | 0.035 | 0.037 | 0.041 |
| acf_mae | 0.117 | 0.104 | 0.171 | 0.128 |
| extremogram_mae | 0.094 | 0.090 | 0.161 | 0.127 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | naive | lstm | cnn | xgboost |
|---|---|---|---|---|
| diff_of_means | 6.056 | -6.790 | -8.562 | 17.315 |
| ratio_of_sd | 0.787 | 0.941 | 0.891 | 0.784 |
| amplitude_ratio_of_means | 0.511 | 0.648 | 0.676 | 0.487 |
| maximum_error | 0.275 | 0.299 | 0.316 | 0.240 |
| ks_mean_on_coarse_res_with_extremes | 0.565 | 0.322 | 0.267 | 0.461 |
| rainy_hours_ratio_of_means | 0.874 | 0.888 | 0.927 | 1.060 |
| qqplot_mae | 0.029 | 0.017 | 0.028 | 0.025 |
| acf_mae | 0.154 | 0.097 | 0.086 | 0.108 |
| extremogram_mae | 0.106 | 0.048 | 0.056 | 0.078 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | lstm | cnn | naive | xgboost |
|---|---|---|---|---|
| diff_of_means | -2.382 | -2.990 | 18.193 | 28.620 |
| ratio_of_sd | 0.902 | 0.849 | 0.734 | 0.728 |
| amplitude_ratio_of_means | 0.601 | 0.634 | 0.465 | 0.427 |
| maximum_error | 0.305 | 0.330 | 0.276 | 0.252 |
| ks_mean_on_coarse_res_with_extremes | 0.288 | 0.215 | 0.473 | 0.409 |
| rainy_hours_ratio_of_means | 0.908 | 0.953 | 0.979 | 1.192 |
| qqplot_mae | 0.026 | 0.038 | 0.036 | 0.040 |
| acf_mae | 0.103 | 0.081 | 0.153 | 0.109 |
| extremogram_mae | 0.084 | 0.062 | 0.135 | 0.106 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97